Download 1 Paper accepted for publication in Language Sciences Explaining

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Esperanto grammar wikipedia , lookup

Modern Greek grammar wikipedia , lookup

Germanic weak verb wikipedia , lookup

Macedonian grammar wikipedia , lookup

Old Norse morphology wikipedia , lookup

Germanic strong verb wikipedia , lookup

French grammar wikipedia , lookup

Inflection wikipedia , lookup

Scottish Gaelic grammar wikipedia , lookup

Ukrainian grammar wikipedia , lookup

Navajo grammar wikipedia , lookup

Old Irish grammar wikipedia , lookup

Japanese grammar wikipedia , lookup

Udmurt grammar wikipedia , lookup

Swedish grammar wikipedia , lookup

Chinese grammar wikipedia , lookup

Modern Hebrew grammar wikipedia , lookup

Kannada grammar wikipedia , lookup

Old English grammar wikipedia , lookup

Turkish grammar wikipedia , lookup

Spanish verbs wikipedia , lookup

Hungarian verbs wikipedia , lookup

Italian grammar wikipedia , lookup

Kagoshima verb conjugations wikipedia , lookup

Lexical semantics wikipedia , lookup

Polish grammar wikipedia , lookup

Ancient Greek verbs wikipedia , lookup

Serbo-Croatian grammar wikipedia , lookup

English clause syntax wikipedia , lookup

Georgian grammar wikipedia , lookup

Portuguese grammar wikipedia , lookup

Spanish grammar wikipedia , lookup

Pipil grammar wikipedia , lookup

Russian grammar wikipedia , lookup

Yiddish grammar wikipedia , lookup

Split infinitive wikipedia , lookup

Ancient Greek grammar wikipedia , lookup

Finnish verb conjugation wikipedia , lookup

Latin syntax wikipedia , lookup

Infinitive wikipedia , lookup

German verbs wikipedia , lookup

Transcript
Paper accepted for publication in Language Sciences
Explaining the distribution of infinitives of impersonals in Russian
In Russian infinitives of impersonal verbs have a peculiar distribution: they are not
acceptable in most syntactic contexts, but there are also syntactic contexts in which they
are perfectly acceptable. Based on a qualitative analysis of data from corpora, the
Internet and an acceptability survey, it is argued that the restrictions on impersonals in
infinitival constructions can be explained if both morphological and semantic-syntactic
factors are taken into account. As is shown, the infinitive in Russian is easily associated
with a human (arbitrary) subject. The restrictions on infinitives can be accounted for in
terms of a semantic-syntactic incompatibility between the meaning of the impersonal
verb, which lacks a subject, and the meaning of the infinitive, which is easily associated
with a human subject. This analysis not only explains the data from Russian, but also
makes predictions about similar data from other languages.
1. Introduction1
As is remarked by Perlmutter and Moore (2002), in Russian there is a restriction on infinitives
of impersonal verbs. Perlmutter and Moore explain this restriction by postulating null (covert,
non-audible or written) forms, more specifically null nominative subjects of finite impersonal
verbs, and the requirement that subjects of infinitival clauses be dative.
1
I benefited greatly from comments by three anonymous reviewers. I am also greatly indebted to Lisa
Cheng, Roberta D’Alessandro and Andries van Helden, for commenting on earlier drafts of this article,
and Boris Kozlov for his help in conducting the acceptability survey.
1
Because of its elegance, Perlmutter and Moore’s explanation of the Russian data is quite
attractive. It can therefore be argued that as long as no alternative explanation is provided, both
the use of null forms and the rule that the case of the subject of the infinitive is dative cannot be
rejected. This opinion is clearly expressed by Perlmutter (2007, p. 304), when he states that
‘[w]hile readers are certainly entitled to their opinions about what is desirable or undesirable, it
is incumbent on those who find null subject undesirable to show that a grammar without them is
superior to one that posits them.’
However, as I will contend in this paper, even though the analysis given by Perlmutter
and Moore (2002) provides an important generalization of the data, it is incomplete, and in
some cases gives an oversimplified picture of the data. Furthermore, the ‘null dative subject
hypothesis’ at times makes wrong predictions. Following Babby (2009), and in contrast to
Perlmutter and Moore (2002), I will contend that data cannot be explained with reference to
case. In contrast to the purely syntactic approach given by both Perlmutter and Moore and
Babby, I will argue that the restriction on infinitives of impersonals is not just syntactic, but also
has a (morpho)semantic dimension.
This article is structured as follows: Section 2 provides an overview of the relevant data;
Section 3 discusses the analysis of infinitives and impersonals in Russian given by Perlmutter
and Moore (2002) and Babby (2009); Section 4 presents an alternative analysis to the same
data; Section 5 provides a general conclusion.
2. The data and theoretical preliminaries
This section sets forth definitions of the terms ‘impersonal verb’ and ‘subject’ and provides a
general overview of the data, as well as the methodology I have used to collect and analyze said
data.
2
For this analysis I will use the term ‘impersonal verb’ with respect to verbs (or
constructions) that do not have a subject that agrees with the verb, and which have third person
singular (neuter) marking when they are used as a finite form, for example:
(1)
Morozilo.
froze-3SG-N
‘It was freezing.’
(2)
Morosilo.
drizzled-3SG-N
‘It was drizzling.’
Weather verbs are typical instances of this type of impersonal (see, e.g. Birjulin (1993) for an
analysis of this class of impersonal verbs), and therefore these meteorological verbs will be my
focus here. However, Russian has a vast array of other types of impersonal verbs and
constructions (see, e.g. Guiraud-Weber (1984) for an overview of such impersonal constructions
in Russian). It should be stressed that some of these impersonal verbs – including
meteorological impersonal verbs – do in fact have personal counterparts with a comparable
meaning that show agreement between the subject and the verb:
(3)
Morosil
doždik.
drizzled-3SG-M rain-N-M
‘The rain was drizzling.’
Furthermore, in some cases, the impersonal nature of the verb is due to the construction, and the
impersonal status of the verb is not part of its argument structure. See, e.g. Babby (2009) for an
overview of the argument structure of various impersonal verbs and impersonal constructions.
3
Note that the class of impersonals under discussion does not include the so-called
‘undetermined personal constructions’ and ‘generalized personal constructions’, which are
traditionally distinguished in Russian grammars (e.g. Švedova 1980). Such constructions are
associated with a non-specific (generic or arbitrary) human first participant, which cannot be
expressed by a nominative form, for example in the following sentences from the Russian
National Corpus (henceforth RNC):
(4)
Stučat. (RNC)
knock-3PL
‘Someone is knocking at the door.’
(5)
Bez
kritiki
ne
without
criticism not
proživeš’. (RNC)
survive-2SG-PRES
‘You (one) cannot survive without criticism.’
These constructions cannot be infinitival, since the specific generic or arbitrary interpretation of
the first participant (subject) is directly related to the second or third person marking on the verb
(for an analysis of similar constructions cross-linguistically within a more formal framework,
see, e.g. Egerland, 2003). As such, they fall outside the scope of the present analysis.
To reiterate, the aforementioned impersonal meteorological verbs will be the focus of this
section, and other types of impersonals will be discussed in section 4.5.
Another term requiring elaboration is ‘subject’. In linguistic literature the term ‘subject’
is defined in various ways, depending on the theoretical framework used by the author. The
present analysis will use the term ‘subject’ in two ways. First, I will use it as a syntactic term for
the form, which agrees with the verb (called ‘morphological subject’ by Babby 2010b).
4
Prototypically this means that the subject is a noun, pronoun or noun phrase with nominative
case marking.2
Second, I will use the term ‘subject’ as a semantic term to refer to the first participant of the
infinitive, that is, the participant with agentive or actor-like properties that is associated with the
infinitive. It should be noted in formal (generative) syntactic frameworks the subject of the
infinitive is also taken to have a particular covert form (PRO). Similarly, such models usually
also regard instances where the first participant of the infinitive is expressed by a dative noun or
pronoun as an actual subject (for example Moore and Perlmutter 2000, and Babby 2009).
I will now turn to the discussion of the data. As remarked above, Perlmutter and Moore
(2002) observe that, in Russian, infinitives of impersonal verbs are not acceptable in most
contexts. There is only one type of context in which infinitives of impersonals are fully
2
I deliberately use the term ‘prototypically’ because there may very well be instances where the form that
governs agreement is not a nominative noun or pronoun. One reviewer argued, for example, that in the
following Russian sentence the prepositional phrase po pjat’ bol’nyx, which is not a nominative noun,
agrees with the verb, which may show either singular or plural morphology:
(i)
U nix v bol’nice ežednevno umirajut/ umiraet po pjat’ bol’nyx.
at them in hospital daily
die.3PL/ die.3SG per five patients-GEN
‘In their hospital five patients die every day.’
In the case of plural agreement one might also give an alternative analysis, and argue that the PP cannot
be seen as the subject, in the same way as in the following sentence from the Internet the nominative
sobaki (‘dogs’) is the subject, and not the genitive plural form štuk (‘pieces’):
(ii)
Zimoj
v Kožuxovskom prijute sobaki umirali po pjat-šest’ štuk
za den’.
winter-INSTR in Kožuxovskij shelter dogs-N died-PL per five-six pieces-GEN a day
‘During winter in the Kožuxovskij shelter dogs were dying five-six pieces a day.’
In the same vein, in (i) the plural agreement of the verb could be explained in the same way as the plural
agreement of the undetermined personal constructions as in (4). Cf.:
(iii)
Na vstreču ot každogo klassa, načinaja s
pjatogo, vybirali
po pjat’ čelovek.
in meeting from every class starting from fifth
selected-3PL per five people
‘When they visited, they selected five people [per class], starting from the fifth class.’ (RNC)
I will refrain from further discussion of the term ‘subject’ here, and refer readers to Babby (2010b) for an
analysis within a formal framework, or Keenan (1976) for a typological approach to subjecthood.
5
acceptable. This is the case if they occur as the complement of auxiliary verbs. An example is
(6) with the auxiliary verb perestat’ (‘stop’), and the infinitive of the impersonal verb morosit’
(‘to drizzle’):
(6)
Perestalo
morosit’. (Perlmutter & Moore 2002, p. 636)
stopped-3SG.N
drizzle-INF
‘It stopped drizzling.’
Such sentences are fully acceptable and absolutely not stylistically marked in any way.
In order to determine in which contexts infinitives of impersonals are not acceptable, I
looked for instances of infinitives of impersonal verbs other than in contexts like (6) in a corpus
of spoken and written Russian (RNC) and on the Internet. In the RNC no examples were
evidenced, however various examples were confirmed on the Internet. Even though the
examples from the Internet were taken from a large variety of sources, ranging from forums to
novels and journals, there are, of course, potential hazards arising from the use of data collected
from the Internet. Because the sample of data is relatively small, it is difficult to determine to
what extent the examples can actually be seen as acceptable instances of modern standard
Russian. Furthermore, in some cases it is not possible to determine who the authors of the
sentences are, and to what extent their speech represents modern standard Russian. To
overcome these difficulties, an acceptability test was conducted with 58 native speakers of
Russian. The participants were asked to grade 62 sentences (including the sentences found on
the Internet) on a five-point scale.3 Participants were asked to rate a sentence as “1” if they
3
The survey included both impersonal meteorological verbs and instances of the accusative experiencer
construction, which will be discussed in section 4.5.
6
found it completely unacceptable, “3” if they were not sure about acceptability and “5” if it was
fully acceptable.4
The survey shows that infinitives of impersonal meteorological verbs are acceptable
with the subordinators pered tem kak (‘before’), and vmesto togo čtoby (‘instead of’), as in the
sentences given below (in each case the mean acceptability score is given, and in some cases the
mean acceptability score of the corresponding finite sentence; see the appendix for the median
scores and standard deviations per sentence):
(7)
Leto
prošlo,
ja èto
čuvstvuju.
Pered tem kak poxolodat’
summer passed
I that
feel
before
get.cold-INF
okončatel’no, vsegda na kakoe-to
vremja
tepleet.5
finally
time
be.warm-3SG.PRES
always for some
‘The summer has ended, I can feel it. Before it finally gets cold, it’s always warm for
some time.’
[mean acceptability 3.53; mean corresponding finite sentence 4.19]
(8)
Solnce
stalo
skatyvat’sja
sun-NOM began go.down-INF
poxolodat’,
stalo
k gorizontu.
No, vmesto togo čtoby
to horizon
but instead.of
žarče.6
4
All the participants are native speakers of Russian (both male and female), mainly between the ages of
20 and 35 years old and having a university education. – The survey was conducted via the Internet.
Participants were sent the file containing the survey question, the sentences and grading scale for each
sentence. The answers were returned by e-mail. The participants were partly found with the help of the
Russian social network “Be in contact” (http://vkontakte.ru). A copy of a test was uploaded on one of the
Russian forums (http://talks.mark-itt.ru/forummisc/blog/6906/8095.html). In addition, people from
Moscow, St. Petersburg, Yekaterinburg, Kazan and Izhevsk were asked to participate in the test using
Internet communication programs such as ICQ and Skype. – The participants were not aware of the
purpose of survey, and the survey was designed in such a way to disguise the intent of the research (e.g.
with respect to the order of the sentences that had to be rated). The participants received a standard
introduction to the survey of about five minutes. The participants filled in the survey while online so that
they could ask questions about the survey if needed. On average it took them ten to fifteen minutes to fill
in the survey. – A number of clearly ungrammatical sentences were added to the survey in order to test
whether the participants filled in the test seriously and attentively. Participants that chose 3, 4 or 5 on one
of the ungrammatical sentences were removed from the test (of the 61 participants, three were removed).
5
<http://erofeeva.livejournal.com/tag/кино>
7
cool.down-INF
become-3SG.N.PAST
warmer
‘The sun started to slide towards the horizon. But instead of getting colder, it became
warmer.’
[mean acceptability 3.31; mean corresponding finite sentence 2.55]
The mean acceptability of these sentences lies between 3 (‘uncertain about the acceptability)
and 4 (‘quite acceptable’) on a 5-point scale, and the median of such sentences is 4. The data
show that there is variability between speakers, which suggests that the acceptability of such
sentences is also partly a matter of style and register. Also note that in the case of (8), with
vmesto togo čtoby, the infinitival sentence is considered to be more acceptable on average than
the corresponding finite version, whereas in the case of (7), with pered tem kak, the
corresponding finite sentence is seen as more acceptable.
It should be noted that in all the examples with vmesto togo čtoby and pered tem kak
substantiated here, both the subordinate clause and the main clause contain a finite
(meteorological) impersonal verb. Apparently, the occurrence of a finite impersonal verb in the
main clause is necessary for the occurrence of an infinitive of an impersonal verb in the
subordinate clause.
Regarding examples on the Internet, none or extremely few were verified of infinitives
of impersonal meteorological verbs with the following subordinators:
-
esli ‘if’;
-
purpose (instrumental) clauses with čtoby;
-
čtoby in the construction of degree (with sliškom/dostatočno X čtoby Y ‘too/enough X to
Y’);
6
<www.gramotey.com/?page=45&open_file>
8
In the survey the following construed sentences were tested, which all show that the
acceptability is lower than 3:
(9)
* Čtoby
in.order
morozit’
na Gavajax, nado,
čtoby
zemlja
freeze-INF
in
that
earth-NOM turn.over-PST
Hawaii,
need
perevernulas’.
[corresponding sentence in English, according to Perlmutter and Moore: ‘In order for it
to freeze in Hawaii, the world would have to turn upside down.’] (Perlmutter & Moore,
2002, p. 621)
[mean acceptability 1.79; mean acceptability corresponding finite sentence 3.17]
(10)
Esli
morozit’,
to
budet očen’
if
freeze-INF.IPFV
then will.be very
xolodno.
cold
[intended meaning: ‘If it is to freeze, it will be very cold.’]
[mean acceptability 2.60]
(11)
V Afrike
sliškom žarko, čtoby
in Africa
too
hot
morozit’.
in.order freeze-INF
[intended meaning: ‘In Africa it’s too hot to freeze.’]
[mean acceptability 2.16; mean acceptability corresponding finite sentence 2.98]
It can, therefore, be concluded that infinitives of meteorological impersonals are acceptable with
a limited number of subordinators. In such sentences both the subordinate clause and the main
clause contain an impersonal verb.
3. Existing analyses
In their paper Perlmutter and Moore (2002) (henceforth also P&M) provide a syntactic analysis
of the distribution of infinitives of impersonals in Russian. They provide example (9) with the
infinitive of the impersonal verb morozit’ (‘to freeze’), which they argue is not acceptable, and
9
its finite counterpart (12), which they argue is fully grammatical. Their acceptability judgments
are confirmed by the acceptability test I have conducted. Sentence (9) is clearly ungrammatical,
with a score below 2, whereas (12) has a mean acceptability above 3:
(12)
Čtoby
morozilo
in.order freeze-3SG-N-PST
na Gavajax, nado čtoby
zemlja
in Hawaii,
earth-NOM turn.over-PST
need that
perevernulas’.
‘In order that it freeze in Hawaii, the world would have to turn upside down.’
[mean acceptability 3.17]
It should be emphasized that the restriction on the infinitive in (9) is due to the impersonal status
of the lexical verb morozit’, since the same construction is perfectly acceptable with an
infinitive of a personal verb, as in (13) with the personal transitive verb issledovat’
(‘investigate’):
(13)
Čtoby
issledovat’
in.order investigate-INF
ètot vopros,
on dolžen
znat’
nemeckij jazyk.
this question
he ought
know-INF German language
(P&M, 2002, p. 635)
‘In order to investigate this question, he should know German.’
P&M explain the restriction on infinitives of impersonals on the basis of a set of rules that are
specific to Russian. First, they argue that surface subjects (i.e. overt subjects) of finite clauses
are nominative (P&M, 2002, p. 621). The next step in the argumentation of P&M is that the
nominative subject is also present when the subject is silent, as in (1) or (2) (P&M, 2002, pp.
630-631). In such sentences, P&M argue, Russian has a silent expletive subject comparable to
English it (the term ‘expletive’ means that the form does not refer to anything).
10
The rule that the subject of a finite verb is nominative is generally accepted in linguistic
literature, but in their argument P&M also provide systematic evidence for another rule, namely
that the subject of the infinitive in Russian is dative. To illustrate this they mention two types of
constructions with a dative infinitive subject, specifically sentences with a conjunction
(subordinator) and an infinitive predicate, and sentences without conjunction (subordinator)
where the infinitive is the main predicate of the sentence. Examples of these constructions with
an overt dative subject are given below:
(14)
[čtoby
nam
uexat’
na vokzal] (P&M, 2002, p. 621)
in.order
us-DAT go.out-INF
to railway-station
‘in order for us to go out to the railway station’
(15)
Mne
ne
sdat’
me-DAT NEG pass-INF
èkzamen. (P&M, 2002, p. 620)
exam-ACC
‘It’s not (in the cards) for me to pass the exam.’
In (14) we find a purposive clause with the conjunction čtoby (‘in order’) and dative pronominal
subject. In (15) we find a so-called infinitival root clause (also called ‘infinitive sentence’, or
‘dative-infinitive construction’), also with a dative pronominal subject. P&M focus on the latter
construction in their discussion. An important attribute of P&M’s explanation involves pointing
out that, while pronoun-drop (pro-drop) of the subject is possible in finite clauses, and even
obligatory in the case of nominative expletives, it does not occur in INFINITIVAL clauses such as
(15). Therefore, the dative subject cannot be omitted in (15), or, as will be explained below, at
least not while maintaining the same meaning:
(16)
*Ne
sdat’
èkzamen. (P&M, 2002, p. 620)
NEG
pass-INF
exam-ACC
P&M’s translation: ‘It’s not (in the cards) for me to pass the exam.’
11
According to P&M (2002, p. 633), this can be seen as evidence that there are no phonologically
null dative pronouns. This provides the solution to the question of why (9) is not acceptable: the
impersonal morozit’ (‘freeze’) requires a silent NOMINATIVE subject, whereas the subject of the
infinitive is DATIVE. Expressed differently, there is an incompatibility of case in (9). Of course,
one could imagine a rule that makes it possible for the silent subject of the impersonal verb to
be assigned a silent dative case. However, P&M argue, pro-drop is not possible in the case of
the dative subject.
Strong evidence in favour of P&M’s claim is that the two types of syntactic contexts in
which one can argue that the case of the infinitive subject is not dative do indeed allow for
infinitives of impersonals. P&M (2002, p. 621, note 2) point out that these two types of contexts
are so-called obligatory control contexts, where the embedded subject often takes the case of the
controller, and so-called raising contexts, where the subject of the infinitival complement bears
the case of the raise.7 This can be exemplified with the following sentences:
(17)
[čtoby
uexat’
na vokzal
in.order
go.out-INF to
railway-station
odnomu]
alone-DAT
‘in order to go out to the railway station alone’
(18)
Ja
xoču
rabotat’
I-NOM want-1SG work-INF
odin.
alone-NOM-SG-MASC
‘I want to work alone.’
(19)
On
načal
he-NOM started-1SG-MASC
rabotat’
odin.
work-INF
alone-NOM-SG-MASC
‘He started to work alone.’
7
It should be noted that not all formal (generative) models make a distinction between raising and
control. See, e.g. Babby (2009) for an alternative analysis of the data, which does not make use of the
notion of interclausal case agreement.
12
According to formal syntactic frameworks (see, e.g. Comrie, 1974; Franks, 1995), in (17) the
form odin (‘alone’) is dative because it agrees in case with the subject of the infinitive, which is
dative by default. However, in control context (18) and raising context (19), the form odin
(‘alone’) is not dative but the same as the case of the nominative subject of the finite verb. As
such, the case assignment to the forms odin (‘alone’) and sam (‘self’) can be seen as a
diagnostic for the presence or absence of a dative infinitive subject. These constructions are
indeed an exception to the rule that infinitives of impersonals are blocked. An example is (20),
with perestat’ (‘stop’) and the infinitive of the impersonal verb morosit’ (‘to drizzle’):
(20)
Perestalo
morosit’. (P&M, 2002, p. 636)
stopped-3SG.N
drizzle-INF
‘It stopped drizzling.’
The data presented by P&M therefore validate their statement that the ‘true generalization is
that they [impersonals] can occur in environments where the subject must be nominative.’
(P&M, 2002, p. 639)
Even though P&M certainly point at an important generalization of the data, their
explanation in terms of covert dative infinitival subjects is not as elegant as it initially seems.
First of all, P&M’s (2002) postulation of dative infinitive subjects does not easily explain why
the restriction on infinitives only occurs in the case of impersonal verbs that are associated with
non-expressed nominative subjects. To return to an earlier example by P&M, why is (13), with
an infinitive of a personal verb, acceptable, whereas a similar construction with an impersonal
verb, such as (9), is not? And why is there a dative subject in (14), even though there is no overt
dative subject in (13)? This can also be illustrated by means of the temporal clauses in (21)
13
(without dative subject) and (22) (with dative subject) (cf. the example provided by P&M, 2002,
p. 621): 8
(21)
Prišlos’
pobrit’sja
pered tem kak
necessary-PRED-PST shave.oneself-INF before
idti
fotografirovat’sja. (RNC)
go-INF having.picture.taken-INF
‘I had to shave myself before having my picture taken.’
(22)
[Tut nastal moj čered.]
Pered tem kak
mne
vystupat’,
before
I-DAT appear-INF
vdrug
podnjalsja
suddenly occurred
šepot.
whisper
‘Then it was my turn. Before I had to go onstage, suddenly people started to whisper.’
(RNC)
P&M suggest a solution in terms of the notion of control. With regard to sentences like (13) or
(23), they argue that ‘in frameworks where controlled subjects are represented as PRO, such
clauses as well are grammatical infinitival clauses with silent subjects’ (2002, p. 635). It is not
8
The use of the dative subject can be seen as a special linguistic device with a special function. Sentences
with a dative express that the situation is ‘planned’, or ‘going to happen’, whereas this modal feature is
not explicitly transmitted in sentences without dative (see Fortuin, 2006, for a detailed analysis how the
modality comes about). This does not accord with Rubinstein (1986), who argues that the main factor
triggering the use of a dative subject in similar sentences with čtoby ‘in order’ is avoidance of coreference
disturbance. Examples where the dative subject is coreferent with the subject in the main clause do,
however, occur, in sentences with a conjunction, e.g.: A pered tem kak emu zaxvorat’, on oslab vdrug do
nevozmožnosti (M. Zoščenko, cited in Fortuin, 2006), ‘But before getting ill, he suddenly became
extremely weak (but before he-dat get.ill-inf.pfv, he got.weak suddenly to impossibility)’. Furthermore,
Fortuin (2006) provides statistical data showing that the percentage of coreferent contexts in sentences
with a conjunction of anteriority, a dative subject and an infinitive clause is not different from sentences
with a nominative subject and a finite predicate. Nevertheless, I am reluctant to deny that coreference (or
emphasis) may be an important trigger for the use of the dative subject. However, this does not exclude
an analysis, which attributes a ‘modal’ meaning to the expression of the dative. It is only in contexts that
are somehow compatible with such a modal meaning that the dative can be used to avoid coreference
disturbance. The association with a modal meaning probably also explains why the dative subject is not
used with all conjunctions of anteriority; it is, for example, very restricted with the conjunction prežde
čem (‘before’) (see Fortuin, 2006).
14
clear to me, however, how this analysis can deal with the fact that in such presumed control
contexts, PRO is associated with dative case. Note, for example, that in (13) the subject of the
infinitive must be dative, which becomes evident if we insert the form sam ‘self’, as in the
following sentence:
(23)
[Čtoby
issledovat’
in order investigate-INF
ètot
vopros
samomu]
this
question
self-DAT
‘In order to investigate this question himself.’
Thus, whereas P&M argue that in controlled constructions we do not have a dative subject but a
silent infinitive subject (PRO), the data clearly suggest that in such contexts the underlying case
of the infinitive subject is dative.
Another problematic issue is that infinitive sentences like (15) do in fact occur without
overt dative subject. In such sentences the infinitive subject is interpreted as a generic or
arbitrary subject, or as the speaker or addressee:
(24)
Pokurit’
by.
Smoke-INF
IRR
‘If only I could have a smoke.’
(25)
Èkzamen ne sdat’.
exam
not pass-INF
‘One cannot pass that exam.’
These interpretations of the subject are due to the absence of a dative noun, and absent if a
dative form is used in these sentences (see Fortuin, 2005, for a more thorough analysis of such
sentences, the question of how the modality comes about, and the semantic and syntactic
conditions for a generic reading, such as the presence of negation or the subjunctive particle by).
15
Again P&M maintain their analysis by arguing that the subject in such sentences must be
analyzed as PRO, and that one cannot speak of pro-drop (2002, p. 633, note 21; p. 634, note
24). 9 Even though the analyses of these problematic cases make sense within the model of
P&M, they also have the character of an ad hoc explanation with respect to the explanation of
the restrictions in impersonals. Furthermore, as I have already shown, their explanation in terms
of case assignment is not able to deal with the entirety of data. It does not explain, for example,
that in some contexts, which are not predicted by the model of P&M, sentences with infinitives
of impersonals are in fact acceptable, as in (7) and (8).
The analysis given by P&M is based on the assumption that impersonals can occur in
environments where the subject must be nominative, and not in contexts where the expressed or
non-expressed case of the infinitive subject is dative. Because of this, P&M contend that the
dative case is in fact the CAUSE of the restriction on impersonals. However, the data suggest that
those contexts where a dative case occurs with the infinitive have something in common, and
that this common feature also explains why impersonals are banned in the majority of the same
contexts.
3.1. Babby’s analysis
Within formal linguistics, an alternative analysis of the data is proposed by Babby (2009).
Babby presents a theory of syntax in which argument structure, which he terms diathesis, plays
an essential part. Babby discusses Russian infinitives in terms of their formation: ‘Infinitive
formation is a diathesis-based operation that composes the diathesis of a lexical verb stem V,
9
A dative pronoun or noun can be omitted in sentences where the modal element of the construction is
expressed by an overt modal form, such as the non-verbal predicate nado (‘necessary’). Cases like these
are not explicitly discussed by P&M (2002), but in an earlier paper Moore and Perlmutter (2000) argue
that the dative cannot be seen as an actual dative subject in sentences with a non-verbal predicate.
Because of this, they behave differently. P&M refer to this in a footnote, where they argue that ‘the dative
nominal in these constructions is not the surface subject’ (2002, p. 623). Similarly, in note 28 (P&M,
2002, p. 637) on the same type of construction, they argue that a dative noun is not a subject and refer to
Moore & Perlmutter (2000) for justification.
16
which is common to all finite and nonfinite verbal categories, with the diathesis of the
infinitive-forming suffix –inf (…). This entails that an infinitive’s syntactic projection consists
of VP embedded as the complement of –inf, which heads its own affixal projection, the
infinitive phrase (…).’ (Babby, 2009, p. 172) According to Babby, an infinitive’s final argument
structure (diathesis) can project to syntax in different ways. With respect to the topic under
discussion, the following syntactic structures are relevant:
(a) Infinitive small clause (s-clause)
(b) Bare infinitive phrase
The s-clause is exemplified by the dative infinitive construction, as in (15), sentences with a
subordinator (complementizer) and an infinitive predicate, as in (13), and a number of other
constructions that all share a common aspect: the case of the (covert or overt) infinitive subject
(PRO) is dative. The diagnostic for this is the occurrence of the forms sam (‘self’) and odin
(‘alone’) in the dative case when they occur with the infinitive. Even though Babby (2009) does
not explicitly discuss this topic, the model he provides explains quite straightforwardly why
infinitive small clauses cannot be impersonal: the infinitive small-clause has a (dative) subject,
whereas impersonal verbs, such as morozit’ (‘freeze’), are subjectless.
The bare infinitive phrase only occurs as the complement of an auxiliary, with which it
forms a complex predicate, as in (20). Impersonal (subjectless) verbs can be infinitives in this
construction because the auxiliary verb inherits the impersonal verb’s external argument. In
generative theory the external argument of a predicate X is the argument that is not contained in
the maximal projection of X. In most cases, this is the subject of the predicate. In the case of
impersonal verbs, there is no external argument; hence the auxiliary inherits the impersonal
status of the infinitive. This is rephrased, as follows, in Babby (2010, p. [16]): ‘The external
argument X of the lexical verb’s diathesis becomes the external argument of the auxiliary verb
and the rest of the lexical verb’s diathesis W becomes the infinitive complement of the
17
auxiliary. The infinitive complement thus gives the impression of controlling its matrix
auxiliary verb because the auxiliary in effect inherits the lexical infinitive’s external argument.’
(see also Babby, 2009, p. 223, for a description in terms of Babby’s theory of diathesis).
In my view, Babby’s analysis of impersonals is more plausible than Perlmutter and
Moore’s analysis, and both theoretically and empirically appealing. However, there are still a
number of questions that are left unanswered. First, Babby’s analysis does not explain or predict
why infinitives of impersonal verbs are possible with some subordinators in Russian. Second,
Babby’s (2009, pp. 220-227) explanation of the restriction on infinitives of impersonals, and the
possibility of such infinitives with auxiliaries, is framed within purely syntactic terms. It would
be possible to strengthen this explanation, I believe, by adding a semantic dimension. In the
remainder of this article such an analysis will be provided. It should be remarked that, even
though my analysis is not framed within a formal syntactic framework, it is not incompatible
with such frameworks.
4. A morpho-syntactic semantic approach
4.1. Infinitives, subjects and agentivity
As is argued in Fortuin (2003), the infinitive denotes in all its different uses a SITUATION TYPE,
and can be seen as a verb (cf. Ebeling, 1984). The term ‘situation’ is used as a cover term for
events, states, processes, etc. Situation types can be seen as abstractions from individual
occurrences of situations that are grouped together on the basis of similarity. Because the
infinitive denotes a situation, it can evoke the thought of a subject of the situation (PRO) and the
idea of the realization of a situation. In contrast to individual situations, which are expressed by
finite verbs, the infinitive does not express person, number, or tense.
There are two reasons why there are restrictions on infinitives of impersonals and not on
finite impersonal verbs. First, infinitives do not express their impersonal status by
18
morphological marking for (third singular) person and (neuter) gender. They can only acquire
an impersonal interpretation if there is another impersonal finite verb of the same type present in
the syntactic context, as in (6), (7) and (8). Second, an important factor triggering the restriction
on infinitives of impersonal verbs is that, without context, the default interpretation of the
infinitive is as having an animate and generic or arbitrary subject, which obviously clashes with
the meaning and status of impersonals. This association can be overruled by establishing
coreference with another subject. In the case of sentences such as (18) or (19), the subject of the
infinitive is coreferent with the subject of the main clause (so-called ‘subject control’), but there
are also sentences where the subject of the infinitive is coreferent with the object of the matrix
verb:
(26)
On
zastavil menja
he-NOM
forced
rabotat’.
me-ACC work-INF
‘He forced me to work.’
A similar pattern can be observed in sentences consisting of a subordinate clause (introduced by
a subordinator), and a main clause. In such sentences the subject of the infinitive is coreferent
with the first participant of the main clause, as in (27)10, or, in the case of object-controlled
infinitives, with the (direct or indirect) object of the matrix verb of the main clause, as in (28):
(27)
Čtoby
pet’,
in.order sing-INF
emu
nužno
oščuščat’ sebja
junošej. (RNC)
he-DAT necessary-PRED feel-INF REFL youth-INSTR
‘In order to sing, he needs to feel young.’
10
I use the term ‘first participant’ as a semantic cover term for what is often called logical subject. The
first participant can be expressed by a nominative noun or pronoun, or by a dative noun or pronoun, and
even by other types of nouns, such as accusative nouns or pronouns. In the following example, the first
participant is expressed by the accusative pronoun of the impersonal experiencer verb tošnit’: Pered tem
kak umeret’, ego tošnilo. (Before die-inf, he-acc nauseate-3sg.n), ‘Before he died, he threw up.’
19
(28)
Ona
predložila Lent’evu
nateret’sja odnim
iz ètix
preparatov
she- NOM suggested Lent’ev- DAT anoint- INF one- INSTR of these preparations
pered tem kak idti
v
parnuju. (RNC)
before
in
steam.room
go-INF
‘She suggested Lent’ev to anoint his body with one of these preparations, before going
into the steam room.’
If no coreference can be established, the infinitive subject is interpreted as a human and generic
(arbitrary) subject (‘one’, ‘people in general’), for example:
(29)
Čtoby
razorvat’
in.order tear-INF
kusok provoloki, trebujutsja
part
wire-GEN
značitel’nye
require-3PL-PRS significant-PL-NOM
usilija. (RNC)
strengths-PL-NOM
‘In order to tear a part of the wire, significant effort is required.’
See, e.g. Ebeling (1984) for a functional-structuralist analysis of the rules of coreference in
Russian infinitives. In addition, note that similar rules are also stated within formal frameworks
using generally accepted notions, such as PRO-licensing and arbitrary PRO.
The association with a human subject in the case of the infinitive can also be found in a
language such as English. For example, consider sentence (9), given by Perlmutter and Moore.
In this sentence the subject of the subordinate clause is an impersonal weather verb (in their
model, the silent subject of morozit’), but the main clause does not contain an impersonal
weather verb. Instead the main clause consists of a non-verbal predicate (nado) and a
subordinate clause introduced by čtoby. The subject of this clause is zemlja. Because there is no
impersonal finite verb available in the main clause, the infinitive subject gets a generic
(arbitrary) reading, which clashes with the meaning of the impersonal verb (‘If one wants to
20
freeze in Hawaii, it is necessary that the world would turn over.’). Note that the absence of an
impersonal verb in the main clause also explains why in the corresponding English sentence
coreference can only be established by using the for it construction in the subordinate clause.
There are data that suggest that the association with a human subject in the case of the infinitive
is stronger in Russian than in a language such as English, even though in English we find a
similar phenomenon. More specifically, in Russian the same construction (čtoby + Xinf + nado
čtoby Y) in fact also prevents a correct interpretation in the following personal sentence, where
the intended meaning is that the subject of the infinitive is coreferent with the INANIMATE
subject of the subordinate clause (‘planet earth’):
(30)
?
Čtoby
rastekat’sja,
in.order crack-INF,
nado, čtoby
zemlja
nakalilas’.
need that
earth-NOM heat-PST
[corresponding sentence in English: ‘In order to split open, it’s necessary that the earth
heats up.’]
As will be argued in the following sections, because the infinitive presupposes the idea of a
subject (first participant), which has to be contextually provided, and because the default
interpretation of the infinitive subject is an animate (generic, arbitrary), infinitives of
meteorological impersonals are not acceptable or marked in most contexts.
4.2. Complement of an auxiliary verb: Infinitives accepted
As mentioned previously, there is only one syntactic context in which there are no restrictions
on the use of an infinitive of an impersonal verb, namely in sentences where the infinitive
functions as the complement of an auxiliary verb. As has also been observed, infinitive
complements of auxiliary verbs differ syntactically from other infinitives in Russian. The main
reason to assume this difference is the nominative case assignment of sam and odin in sentences
21
such as (18) and (19). Based on earlier observations by Comrie (1974), Franks (1990) argues
that the nominative subject is assigned in those contexts where the infinitive can be seen as a
verbal phrase (VP), whereas in contexts where the dative subject is assigned, the infinitive
functions as a complementizer phrase (CP) or clause if a complementizer is absent. That the
infinitive and the auxiliary form a complex predicate is an important feature of such a VP. A
similar view can be found in Babby (2009), as discussed in the previous section, who also
asserts that infinitive complements of auxiliary verbs must be analyzed as infinitive phrases,
which form a complex predicate with the auxiliary verb.
In my view, the syntactic analyses mentioned here can certainly be complemented with
a semantic dimension (see also Fortuin, 2003). Auxiliary verbs indicate a phase of a particular
situation (e.g. ‘to begin X’) or a modal relation with a particular situation (‘to want to X’). The
situation expressed by the auxiliary not only forms a complex predicate with the infinitive
syntactically, but also semantically, since they refer to a complex single event. This complex
event conceptual structure explains why the auxiliary verb can inherit the impersonal status of
the infinitival verb as described by Babby (2009). Phrased differently, the conceptual structure
(‘complex event’) is reflected in the syntactic structure with an IMPERSONALLY MARKED
auxiliary, which forms a complex predicate with the impersonal infinitive. As a result the
infinitive is not associated with agentive features.
4.3. Infinitives occurring with a subordinator
As I have discussed above, not all subordinators are created equally with respect to the use of
infinitives of impersonals: with some subordinators, infinitives of impersonals are acceptable,
whereas with others they are excluded. The data can be summarized as follows:
Infinitives of meteorological impersonals are acceptable:
22
•
Subordinator pered tem kak (‘just before’);
•
Subordinator vmesto togo čtoby (‘instead of’);
Infinitives of meteorological impersonals not acceptable:
•
Subordinators of purpose with čtoby;
•
Subordinator esli (‘if’);
•
Sentences with sliškom/dostatočno and the subordinator čtoby;
As I will argue, the reason why some subordinators are fully incompatible with impersonal
infinitives is that they presuppose the feature of [+agent]. This means they presuppose a human,
usually intentional first participant of the situation expressed by the predicate. The clearest
example of such a subordinator is esli. The conjunction esli can be combined with an infinitive
in sentences such as the following:
(31)
Esli vstavat’
rano, uvidiš’
rassvet. (Bricyn, 1990)
if
early see-2SG.FUT.PFV
sunrise
get.up-INF
‘If one is to get up early, one will see the sun rise.’
According to Bricyn (1990), in sentences with [esli ‘ if’ + infinitive], the subject of the infinitive
is obligatorily animate and generic (‘you’, ‘one’, ‘people in general including the speaker’).
This explains why sentences such as the following, where the subject of the infinitive is
coreferent with a third person subject in the main clause, and not generic, are not acceptable11:
11
The construction is possible with first person subjects in the main clause if the infinitive subject can be
interpreted as a generic subject:
(1)
Esli govorit’ prosto, to my stremimsja byt’ lučšim universitetom v mire (…).
if
talk-INF simply, then we try
be-INF best
university
in world
23
(32)
* Esli vstavat’
if
get.up-INF.IPFV
rano, on uvidit
rassvet. (Bricyn, 1990)
early he see-3SG.FUT.PFV
sunrise
‘If he is to get up early, he will see the sun rise.’
[mean acceptability 2.43]
I have not found any examples of sentences like (32) in the RNC and on the Internet.
Furthermore, in our survey the mean acceptability of (32) was under 3 (‘not sure about the
acceptability), which corroborates the analysis by Bricyn (1990) that the subject of the infinitive
of infinitival esli clauses can only be generic. The same rule also explains why sentences with
esli cannot occur with non-animate subjects or weather verbs, because the subjects of such verbs
cannot be associated with a generic subject:12
(33)
* Esli morozit’,
if
freeze-INF.IPFV
to
budet očen’
then will.be very
xolodno.
cold
‘If it is to freeze, it will be very cold.’
[mean acceptability 2.60]
Similarly, the construction [čtoby (‘in order to’) + Xinfinitive][nado, čtoby Y (‘it is necessary that
Y’)], as in (9), cannot be combined with an impersonal infinitive, because it presupposes a
human intentional agent. The meaning of this construction can be paraphrased with ‘If (a
person) strives for X, then Y is necessary’ (cf. von Fintel & Iatridou, 2007). Obviously, this
‘If you want to put it simply, we try to be the best university in the world.’ (RNC)
12
Note that the infinitive subject may get a non-generic reading or occur with non-animate subjects if the
subject of the infinitive is associated with a dative noun in the subordinate clause. However, since
meteorological impersonal verbs have no subject that can occur in the dative, they cannot occur in this
construction.
24
purposive and instrumental meaning is not in accordance with the semantics of an impersonal
meteorological verb.
The feature [+agent] also plays a part in the case of the construction with the degree
expression sliškom ‘too’ or dostatočno ‘enough’, and a subordinate clause introduced by čtoby,
as in (11). This construction has a different semantic-syntactic structure than the other two
constructions with a subordinator discussed above. In contrast to those, the construction is a
comparative-like construction. The subordinate clause is not an inherent or obligatory element
of the construction, but expresses the consequences of the degree expressed in the main clause.
In the case of the construction of degree, the consequences expressed by the subordinate clause
have a modal character. More specifically, the subordinate clause presupposes a human agent
that is able or unable to do something. Informally, the construction [dostatočno X, to Y] means
that the degree (or quantity of) X is such that Y can be realized. From my standpoint, this modal
character of the construction explains why it is not possible to use an infinitive of an impersonal
in this particular construction.
I will now turn to the subordinators that can be combined with infinitives of
impersonals, namely sentences with vmesto togo čtoby (‘instead of’), and pered tem kak (‘just
before’). The meaning of [vmesto togo čtoby XINF, Y] can roughly be paraphrased as ‘one could
think that X would occur, but Y occurred’. This subordinator does not presuppose the semantic
feature [+agent], because of which it is perfectly in accordance with an inanimate subject or
impersonal verb. Furthermore, the data from the corpus (RNC) show that the large majority of
sentences with this conjunction have an infinitive predicate, and that the use of a finite predicate
is very infrequent. It may be that this factor further contributes to the preference for infinitives,
even in the case of impersonal weather verbs.
Similarly, the subordinator pered tem kak does not presuppose the feature [+agent], but
merely indicates that the situation described in the subordinate clause occurs just before the
situation expressed in the main clause. It should be noted here that even though the use of
infinitives of impersonal weather verbs is possible with pered tem kak, the finite counterpart is
25
preferred. This was illustrated earlier with example (7), which had a mean acceptability of 3.53
in our survey, whereas its finite counterpart had a mean acceptability of 4.19. Data from the
Internet indicate a corresponding tendency. A search on the Internet (via Google and Yandex)
for the conjunctions pered tem kak, and do togo kak and the impersonal verbs (po)xolodat’ (‘get
cold’), morozit’ (‘freeze’), and smerkat’sja/smerknut’sja (‘get dark’) resulted in five instances
with an infinitive, all with pered tem kak, and 26 instances with a finite verb (eighteen instances
with do togo kak; eight instances with pered tem kak).13 Even though the conjunction pered tem
kak typically occurs with an infinitive predicate (Fortuin, 2006), this suggests more examples
were confirmed with a finite weather verb than with an infinitival weather verb on the Internet.
There must be an explanation accounting for both properties. Why are such sentences possible,
and why are they also restricted? To answer the first question, the conjunction pered tem kak
does not necessarily presuppose the idea of an intentional human agent. As such, the use of
infinitives of impersonals is not fully blocked. The reason why infinitives of impersonal weather
verbs are acceptable, even though in the same context the finite counterpart is preferred, can
probably therefore be attributed to the general restrictions on impersonal infinitives in
constructions with a conjunction and an infinitive clause.
Finally, I will discuss a construction without subordinator that has not been discussed
thus far, one that cannot be combined with infinitives of impersonal verbs. This construction
contains a non-verbal predicate such as nevozmožno (‘impossible’) or trudno (‘difficult’), which
typically expresses a state that can be experienced by people, similar to English constructions
such as It’s difficult to X; It’s time to X, etc. An example of this construction is given below
with the non-verbal predicate trudno (‘difficult’):
(34)
Rabotat’
bylo
trudno.
work-INF
be-3SG.N.PST
difficult-PRED
13
In the case of the verb smerkat’sja, the construction contained a finite form of načat’/načinat’ ‘begin’,
and an impersonal infinitive. These instances are counted as finite.
26
‘It was difficult to work.’
The non-verbal predicate occurs with a neuter singular form of the verb ‘to be’ (byt’) (which is
not expressed in the present tense in Russian). In (34) the subject of the infinitive is either
interpreted as contextually given (e.g. associated with the speaker), or as a generic subject
(‘one’, ‘people in general’). By using a dative noun or pronoun, the subject of the infinitive may
be associated with a specific person who functions as the experiencer of the state expressed by
the predicate:
(35)
Mne
bylo
I-DAT bes-3SG.N.PST
trudno
rabotat’.
difficult-PRED work-INF
‘I found it difficult to work.’
What about the use of impersonal weather verbs in this construction? Since non-verbal
predicates like trudno or nevozmožno express states that can be experienced by people, they
cannot be combined with impersonal weather verbs, for example:
(36)
*
V Afrike nevozmožno
morozit’.
in Africa impossible
freeze-INF
[corresponding sentence in English: ‘In Africa it’s impossible for it to freeze.’]
The construction [nevozmožno + infinitive] expresses internal possibility (ability, capacity),
which presupposes a human agent. This is semantically incompatible with impersonal weather
verbs. As is shown by the translation of (36), a similar restriction also occurs in the English
[impossible + infinitive] construction, where coreference is established by using for it. The
predicate nevozmožno can, however, occur with an impersonal finite verb in sentences that are
introduced with a subordinator (čtoby), as in (37):
27
(37)
Nevozmožno,
čtoby
morozilo v
impossible-PRED that-IRR froze
in
Afrike.
Africa
‘It is impossible that it freezes in Africa.’
Such sentences express a different kind of modality, one that is closer to epistemic modality (‘it
can’t be true that the situation expressed by the infinitive is the case’). For this reason, this
construction is not incompatible with meteorological verbs.14
The only non-verbal predicate that was evidenced with impersonals is the predicate
rano (‘early’), as in (38):
(38)
ešče svetat’,
Rano,
rano
early-PRED
early-PRED still
rabota ne
get.light-INF work
not
sdelana.15
done
‘It is still too early to get light, the work isn’t done yet.’
[mean acceptability 3.57]
A possible explanation is that rano is less strongly associated with a human subject than the
other non-verbal predicates. In other words, the concept expressed by nevozmožno and an
infinitive is associated with a human (expressing internal possibility, ‘being able to’), whereas
rano expresses a point on a scale ranging from ‘early’ to ‘late’ – this concept is not inherently
associated with a human participant. Furthermore, the construction [rano + infinitive] expresses
a relative property (‘too early’), and the infinitive SPECIFIES in what respect this property
applies: (‘early with respect to X’). If a dative is expressed, it does not express a prototypical
experiencer, but rather the person with respect to whom the situation expressed by the predicate
and the infinitive applies:
14
The construction [nevozmožno + čtoby + infinitive] seems to be very restricted in Russian. In the RNC
only two examples were found. These examples seem to express ability rather than epistemic possibility.
15
<www.proza.ru/texts/2004/05/21>
28
(39)
A
mne ešče rano
but I-DAT still
early-PRED
umirat’. (RNC)
die-INF
‘It is still (too) early for me to die.’
As such, it differs from prototypical ‘experiencer’ predicates such as trudno ‘difficult’, as in
(35). With this predicate it is communicated that that the subject expressed by the dative form
experiences the realization of the infinitive situation by him as difficult. Such experiencer
semantics – inherently associated with a human subject – are absent with rano. This makes rano
a more likely candidate to be combined with impersonal weather verbs than other non-verbal
predicates. However, because the [predicate + infinitive] construction is easily associated with
an agentive meaning, sentences like these may easily suggest would-be control over the
circumstance of getting light, as if there might be a human subject somehow responsible for the
fact that it becomes light. As such, they are not stylistically neutral and still demonstrate the
typical agentive semantics associated with the infinitive.
4.4. Infinitives and agentivity from a cross-linguistic perspective
I have argued that the absence of person marking and the association with a human (generic)
subject is one of the main factors that trigger the restriction on impersonals. One could expect to
see similar features across languages, since absence of impersonal marking is a feature of
infinitives in other languages as well. A cross-linguistic analysis falls out of the scope of the
present discussion, but I will make a few remarks and observations.
If we examine Italian, a Romance language, we can perceive both similarities and
differences. Italian impersonal meteorological verbs are comparable to Russian impersonal
meteorological verbs. In both languages, these verbs do not have any arguments in their
argument structure (see also Babby, 2009, p. 261), and they occur without expletive. As in
29
Russian, there are no restrictions on infinitives of impersonal weather verbs if they are used as
complements of an auxiliary. Furthermore, similar to Russian, the use of impersonals with nonverbal predicates like possibile (‘possible’) is not acceptable, because it suggests an agentive
reading:
(40)
* E’ possibile piovere in Africa.
is possible
(41)
rain-INF in Africa
E’ possibile che piova in Africa.
is possible that rains
in Africa
‘It’s possible for it to rain in Africa.’
Of course, the same is true for English, where It’s impossible to rain is not acceptable, either.
Italian differs, however, from Russian with regard to the use of infinitives that occur with
conjunctions. In Italian the following conjunctions can occur with an infinitive predicate: per
(‘in order’/ to’), prima di (‘before’), dopo (+ aver; ‘have’) (‘after’), senza (‘without’), invece di
(‘instead of’), oltre ad (‘besides’), fino a (‘until’). Each of these conjunctions can occur with
infinitives of impersonal weather verbs. Below some examples from the Internet are given:
(42)
Prima di
piovere,
lampeggia.
before
rain-INF
lightens
‘Before it rains, there is lightning.’
(43)
Qui
invece
here instead
di nevicare, piove.
of snow-INF rains
‘Here it rains instead of snowing.’
(44)
Oltre a nevicare sul resto del nord nevica anche in
Liguria?
besides snow-INF in
Liguria
rest
of
north snows also
in
‘Besides snowing in the rest of the North, it also snows in Liguria?’
30
(45)
Inizialmente tuonava
lontano,
poi
First
far.away
afterwards has started
thundered
ha cominciato ad avvicinarsi
sempre di più fino a tuonare
dietro la
always
behind the beach
more
until
thunder-INF
to get.closer-INF
spiaggia.
‘First the thunder was far away, but then it started to approach until there was thunder
behind the beach.’
(46)
Dopo una pausa di circa 1 ora
after
a
break
senza
nevicare sta ricominciando!
of about 1 hour without snow-INF is
starting.again
‘After a break of about one hour without snowing, it started again.’
Like in Russian, both the subordinate clause and the main clause contain an impersonal verb in
the examples given above. Sentences like these, with infinitives of impersonal verbs, seem to be
relatively frequent in Italian. Data from the Internet found using Google suggest that the
infinitive is used even more frequently than its finite counterpart.
Note, however, that the restrictions occurring in Russian are not an idiosyncratic
phenomenon of Russian. Similar restrictions apply, for example, to Dutch. This can be
illustrated with the conjunction door(dat) (‘because’, ‘by’), which can occur both with personal
and impersonal finite verbs, but only with personal infinitives. This explains why (47), with a
finite form of the impersonal meteorological verb vriezen (‘freeze’) and the expletive het (‘it’),
is acceptable, whereas its infinitival counterpart in (48) is not:16
16
As in Russian, there also seem to be differences between impersonal verbs with respect to their use.
Note, for example, that the impersonal transitive verbs duizelen (‘feel dizzy’) or dagen (‘remember’), as
in (1), can only be infinitival if the infinitive functions as the complement of an auxiliary verb, as in (2):
(1)
(2)
Het duizelt
it
make.dizzy-3SG
‘I am feeling dizzy.’
Het begon
me
it
started- 3SG
me
‘I started to feel dizzy.’
me.
me
te
to
duizelen.
make.dizzy-INF
31
(47)
Doordat het vriest, kan het sneeuwen.
because it
freezes, can it
snow-INF
‘Because of the frost, it can snow.’17
(48)
*Door
te vriezen,
kan het sneeuwen.
because to freeze-INF can it
snow-INF
[corresponding sentence in English: ‘Because of the frost, it can snow.’]
These data seem to suggest that there are some cross-linguistic similarities between Russian,
Italian, Dutch and English, but also some differences. In all these languages, including Dutch,
there are no restrictions on infinitives of impersonals if they are used as complements of
auxiliaries (modal and phase verbs). Furthermore, in all these languages, the use of infinitives of
impersonal verbs is unacceptable with non-verbal predicates like ‘possible’. This is also the case
in Dutch; cf. (40), with the following Dutch example:
(49)
* Het is onmogelijk om
it
te regenen in Afrika.
is impossible in.order to rain-INF in Africa
‘It is impossible for it to rain in Africa.’
There are, however, some differences with respect to the use of infinitives with conjunctions.
More specifically, in Italian there seem to be few or at least fewer restrictions on infinitives of
It is not possible to use the infinitive of these verbs with subordinators. As such they differ from
meteorological impersonals, which can be infinitival with some subordinators. This is at least true for the
meteorological verb sneeuwen (‘snow’) and the subordinator in plaats van (‘instead of’) (cf. (48)):
(3)
Toen we daar aankwamen, bleek
het te regenen in plaats van te sneeuwen.
when we there arrived,
turned.out it to rain-INF instead of to snow-INF
‘When we arrived, it turned out to rain, instead of snowing.’
(example taken from the Internet: http://www.klasse.be/leraren/archief/13347)
17
A more literature translation of the Dutch sentence would be ‘Because it freezes, it can snow.’
32
impersonal verbs than in Russian or Dutch in these contexts. The data suggest that Italian
[subordinator + infinitive] constructions are less easily associated with agentive features than
comparable Russian constructions. English differs from both Russian and Italian in having very
few subordinators that can be combined with infinitives. Instead, English uses a gerund with
most subordinators (e.g. instead of, without, before, after). The subordinator (in order) to, which
can be combined with an infinitive, cannot be combined with an infinitive of an impersonal verb
if the verb has a clear purposive meaning (as in the translation of (9), without for it).18
Obviously, in explaining such cross-linguistic similarities and differences, an
explanation like the dative subject hypothesis provided by Perlmutter and Moore (2002) is not
very helpful, since it fully relies on a language-specific factor of Russian (the presence of a
dative subject of the infinitive) for which there is no evidence in English, Italian and Dutch.
How, then, should both the similarities and the differences be accounted for? The most
important similarities can be reduced to the fact that in all four languages, as a general rule, the
identity of the subject of the infinitive must be established by coreference, or is interpreted as a
generic subject. In all four languages, if restrictions on infinitives of impersonal verbs occur,
they are connected to the impossibility of providing the infinitive with an impersonal
interpretation. Furthermore, determining the impersonal status of the infinitive is difficult in
sentences with a non-verbal predicate in all four languages, whereas no such problems arise in
sentences where the infinitive is used as a complement of a finite verb. Constructions where the
infinitive occurs with a subordinator represent a diverse picture, as there are clear differences
between the different languages. The data suggest that in Italian, infinitives are less verbal and
much more action nominals when compared to Russian. As such, Italian infinitives may share
18
I do not want to conceal that there is one difference between English and Russian for which it is
difficult to account. In Russian, infinitives of impersonals are not acceptable with sliškom (‘too’), as in
(11), but in English the corresponding construction seems to be acceptable, for example: Is it ever too
cold to snow? (Example taken from C. D. Ahrens, 2007. Meteorology today: an introduction to weather,
climate, and the environment. Thomson/Brooks/Cole, Belmont, USA, p. 164). A relevant factor that
should be considered is that in Russian the same construction is in fact not very acceptable with a finite
meteorological verb, either (mean acceptability 2.98). This suggests that the construction as such blocks
meteorological impersonal verbs.
33
similarities with categories like gerunds or deverbal nouns. This is in accordance with other
independent evidence. For example, note that in Italian infinitives of impersonal verbs can also
be used with determiners or demonstrative pronouns, forms which can only be combined with
nominal forms:
(50)
questo nevicare
this
rain-INF
‘this raining’
(51)
il
nevicare
the rain-INF
‘the raining’
Russian has no determiners, but the use of demonstrative pronouns with infinitives is
ungrammatical: *èto morozit’ (that freeze-inf). Another feature of Italian infinitives that
indicates a different status is that impersonal infinitives can also be used in contexts where there
is no other impersonal verb present in the context, for example:
(52)
Se continua
l’umido,
senza piovere troppo, potremmo tornare sul Morello.
if continues.3SG the moist, without rain-INF too
could.2PL return on Morello
‘If the moist continues [it continues to be moist], without raining too much, we can
return to the Morello [mountain].
This may perhaps also be linked to the general tendency of Italian (or Romance languages in
general) to use non-finite forms (participles, infinitives, nominalizations) where Germanic
languages, and perhaps also Russian, tend to use finite forms (see, e.g. Korzen 2008). Besides
these factors, there may be other aspects at play as well, such as the status of the temporal
conjunctions in Italian, which can also function as prepositions, e.g. prima della fine (‘before
34
the end’). In order to evaluate whether these factors are relevant, further cross-linguistic
research is needed, including an acceptability survey for the various languages.
4.5. Other impersonal verbs and constructions
Perlmutter and Moore (2002) argue that the restriction on infinitives of impersonals holds for
ALL
impersonal verbs or constructions, not just for meteorological verbs. They illustrate this
with various examples, including examples of the accusative human experiencer construction,
as in (53), which cannot be infinitival, as in (54):
(53)
Menja
tošnit. (2002, p. 628)
me-ACC nauseate-3SG-N
‘I feel nauseous.’
(54)
* [čtoby
[in.order
menja
tošnit’
me-ACC nauseate-INF
zimoj] (2002, p. 628)
winter-INSTR]
‘in order for me to feel nauseous in the winter’
It should be noted, however, that infinitives of accusative impersonal verbs are evidenced on the
Internet, most specifically in sentences with the conjunctions: vmesto togo čtoby and pered tem
kak: 19
19
In the RNC one example was evidenced of a conjunction and infinitive of an accusative experiencer
verb (the verb lixoradit’ ‘be feverish’):
(1)
Vozmožno, čto
ustrojus’
na stabil’nuju,
priličnuju rabotu, s tem, čtoby
possible
that arrange-1SG
for stable
nice
job
so.that
ne lixoradit’
v ožidanii
otveta
iz
Prokuratury. (RNC)
not be.feverish-INF
in anticipation
answer-GEN
from prosecutor’s.office
‘Perhaps I’ll arrange a stable nice job, so that I won’t have to feel feverish while waiting for the
answer from the prosecutor’s office.’
The native speakers I consulted found this sentence rather marked, even though some of them considered
it to be acceptable. One of the native speakers suggested that it is the impossibility to build other
35
(55)
I
vmesto togo čtoby
and instead.of
stošnit’,
menja načalo
korčit’
throw.up-INF
I-ACC began.3SG-N writhe-INF
smexa.20
ot
from laughter
‘And instead of getting sick (throwing up), I started to writhe with laughter.’
[mean acceptability 3.91; mean corresponding finite construction 3.00]
(56)
[P]rimerno raz
about
v polčasa/čas
ee
once in half.our/hour
tošnit.
she-ACC throws.up-3SG
plačet).21
(pered tem kak
stošnit’
ona
before
throw.up-INF
she-NOM cry-3SG-PRS
‘About once every half hour/hour she [= the cat] throws up (before throwing up, she
cries).’
[mean acceptability 3.29]22
Other instances were confirmed on the Internet as well, but these were on average judged to be
unacceptable in our survey, e.g.:
(57)
I
postepenno
and constantly
skatyvaetsja k patetike.
Ee
gravitates
she-GEN not so
to patheticism
ne tak mnogo,
much
constructions with the intended meaning that makes the construction acceptable. Because of this, the
usually impersonal intransitive verb is compelled (“coerced”) to instead become a personal intransitive
verb. An alternative explanation is that lixoradit’ is used as a regular personal intransitive verb. This use
exists in Russian, even though it is archaic (see, e.g. Ožegov, 1972).
20
<asya72.livejournal.com/2006/08/30/>
21
<www.mgks.zooclub.ru/questions.php?idconsult=5&page=5>
22
The finite counterpart of this sentence was not included in the survey. In order to gain some insight into
the difference between the infinitival and the finite construction, we conducted a small-scale test by
asking ten native speakers of Russian to indicate whether they preferred the version with the infinitive or
the version with the impersonal finite verb (pered tem kak ee tošnit ‘before she vomits (before she-acc
vomits-3sg).’ Three out of ten people preferred the infinitive. Note that in this sentence, it is the sentence
preceding the subordinate clause, and not the main clause, that contains a finite impersonal verb.
36
čtoby
stošnit’,
no i
ne tak malo, čtoby
in.order nauseate-INF but also not so
little
ne zametit’.23
in.order not notice-INF
‘And it constantly tends to get pathetic. Not that much as to make one nauseous, but not
that little to not notice it, either.’
[mean acceptability 2.29]
Furthermore, it is probably not a coincidence that all the verified examples of infinitives of
accusative experiencer verbs contain infinitives of (s)tošnit’/(vy)rvat’ (‘throw up’), and that no
examples are evidenced as containing infinitives of (za)znobit’ (‘shivering’). A possible
explanation is because of the presence of different forums where people can pose questions to
doctors about health, and where the topic of ‘vomiting’ is simply a more popular topic than
‘shivering’. One could also hypothesize that there is a semantic reason why verbs expressing
‘throw up’ would behave differently from verbs expressing ‘shivering’. Since čtoby presupposes
a goal-oriented action, it can only be combined with accusative experiencer verbs if the verb can
be interpreted as the indirect result this action. Perhaps, in the case of such verbs, even the use
of the infinitive is (marginally) possible, because the infinitive easily suggests an agentive
subject. In contrast to the idea of ‘throwing up’, the occurrences of ‘shivering’, or ‘having a
fever’ are less easily considered as something semi-controllable. This probably blocks the use of
an infinitive of these verbs altogether, but allows for the use of a finite verb. The relation
between the infinitive and the idea of ‘control’ or ‘agency’ may also explain why in
constructions where čtoby does not indicate a goal, the use of infinitives of verbs like (s)tošnit’
(‘feel sick’, ‘throw up’) is less restricted. Additionally, it is possible that the use of infinitives of
verbs like stošnit’ is facilitated in modern informal Russian because there is a personal transitive
use of this verb. Examples of this use can in fact be verified on the Internet and in the RNC, but
are considered to be ungrammatical according to our survey:
23
<http://ajvideo.narod.ru/AlFIlm/LastSamurai.htm>
37
(58)
Nedeli dve nazad on stošnil
week
two ago
ves’
he threw.up-3SG-M whole-ACC
zavtrak (…).24
breakfast-ACC
‘A week or two ago, he threw up his whole breakfast.’
[mean acceptability 2.55]
As such, sentences like these must be regarded as substandard. In order to test whether people
who accept infinitives of (s)tošnit’ do so because, for them, it is an impersonal verb, it was
investigated whether there is a correlation (Pearson) between the acceptability score that
individuals assign to infinitives of (s)tošnit’ and the score they assign to personal use of
(s)tošnit’ as in (58).25 The results of this test are given in Table 1.
Table 1
Correlation acceptability infinitival of impersonal (s)tošnit’ and finite transitive personal use
Number of
Correlation with acceptability of (58)
sentence
(55)
8% (p = 0,559; not significant at the 0,05 level)
(56)
26% (p = 0, 042)
(57)
43% (p = 0,001)
This test shows that there is indeed a very weak correlation between the rating of the personal
use in (58) and the infinitival use in (56) with pered tem kak and a strong correlation between
24
<www.webapteka.ru/veterinary_arc/.../p34.html>
An argument against this hypothesis is that in (56) we find an infinitive of (s)tošnit’ and an impersonal
present tense use in the same fragment, indicating that people who employ the infinitival form of the
construction also use the regular finite impersonal construction. However, it may be that some people
allow for different versions of the construction: (i) impersonal (intransitive) use, (ii) personal (transitive)
use and (iii) personal (transitive) infinitival use.
25
38
the personal use and the infinitival use in the case of (57). However, there is no significant
correlation with the personal use in (58) in the case of (55) with vmesto togo čtoby. In the latter
case this is because this sentence is considered to be acceptable by people that do not accept
personal transitive use of (s)tošnit’.
These data could be taken as an argument that, in the case of verbs expressing vomiting,
use of the infinitival form is possible at least for some speakers, because in this case the
infinitive is not interpreted as an impersonal but as a personal verb, something that is not in
accordance with normative standard Russian. Note, however, that the correlations given earlier
do not necessarily or strictly point to a relation of cause and effect (people that accept infinitives
of accusative experience verbs do so BECAUSE for them it is a personal usage). Instead, it may
be that some people that accept infinitives of (s)tošnit’ or (vy)rvat’ (partly) do so because they
are liberal with respect to what is acceptable or not acceptable.
At this point it is important to stress that the sentences discussed here are to a large
extent peripheral; this even includes the sentences that were judged to be quite acceptable in the
survey. To give an example, the standard way to express a relation of anteriority between an
event like ‘throwing up’ and another event is the use of a nominalization, instead of an infinitive
or even a finite verb. In the RNC three examples were determined with the preposition pered
(‘before’) and the nominalization rvota (‘vomit’), whereas no examples were found with pered
tem kak and an accusative experiencer verb. Furthermore, combinations such as pered pozyvom
rvoty ‘before the urge to throw up (before urge-instr vomit-gen)’ or pered nastupleniem rvoty
‘before the start of the throwing up (before start-instr vomit-gen)’ can be found quite frequently
on the Internet, and have a high acceptability score, e.g.:
(59)
Pered pozyvom rvoty
before urge
u rebenka podnjalas
vomit-nmlz-gen at child
rose
temperatura.
temperature
‘Before the urge to throw up [occurred], the temperature of the child rose.’
39
[mean acceptability: 4.28; median 5]
It is important to note that with other impersonal verbs or constructions, as discussed by
Perlmutter and Moore (2002), there are no exceptions and the restriction fully applies.
Impersonal sentences with a genitive of negation or a non-verbal predicate, such as (60) and
(61), respectively, were not evidenced on the Internet:
(60)
*[čtoby
[in.order
v prudu ne plavat’
kuvšinok] (P&M, 2002, p. 626)
in pond
water lilies-GEN]
not float-INF
[corresponding meaning in English according to P&M: ‘in order for there not to be
any water lilies floating in the pond’]
(61)
*Vmesto togo čtoby
instead.of,
byt’
grustno,
be-INF sad-PRED,
mne
bylo
veselo.
I-DAT was-3SG-N happy-PRED
[intended meaning in English: ‘Instead of feeling sad, I felt happy.’]
Sentences with the non-verbal predicate žal’, such as the following, were evidenced very
infrequently on the Internet, and were considered non-acceptable in the survey:
(62)
*Čtoby
Borisu
byt’
žal’
sobak (P&M, 2002, p. 628)26
in.order Boris-DAT be-INF sorry-PRED dogs-ACC
[intended meaning in English: ‘in order for Boris to feel sorry about the dogs’]
26
On the site <lyrics.filestube.com> two examples were found with an infinitive of žal’ (‘feel sorry’), in
both cases translations of English song texts. In our survey we tested the following example: Èto sliškom
pozdno čtoby byt’ žal’ ‘It is too late to be sorry (that too late-pred to be-inf sorry-pred).’ This sentence
had a mean acceptability of only 1.28 (ungrammatical).
40
Finally, despite that very few instances of infinitives of reflexive impersonal verbs with a dative
subject were substantiated, such as rabotat’sja ‘to (feel like) working’, the native speakers I
consulted did not consider these sentences acceptable:
(63)
*[čtoby
Borisu
rabotat’sja
in.order Boris-DAT work-REFL-INF
doma]27
at.home
[intended meaning in English according to P&M (2002, p. 628): ‘in order for Boris to
be able to work at home’].
One can hypothesize that meteorological verbs behave differently from other impersonal verbs
or constructions because their impersonal character is an inherent part of their argument
structure (other than, for example, the genitive of negation construction). Moreover, they are not
associated with experiencer semantics, as in the case accusative experiencer verbs or impersonal
reflexive verbs, which usually require the expression of a dative or accusative form. To what
extent it is possible to provide a general explanation for all impersonal verbs, or whether each
verb or construction has to be explained differently, is something that falls beyond the scope of
this analysis.
5. Conclusion and further remarks
27
It should be noted that, on the Internet, instances with čtoby (‘(in order) to’) and an infinitive of
rabotat’sja (work-refl) are evidenced, for example in the construction of degree: Material ètix trub
obyčno byl (…) dostatočno mjagkim, čtoby rabotat’sja takimi instrumentami.
<www.smichok.ru/tag/kotoroe/ > ‘The material of those pipes was usually soft enough to be able to work
well with such instruments (material-nom those-gen pipes-gen usually was enough soft-adj to work-refl
such-instr instruments-instr).’ However, since the same verb is also used as a personal verb on the
Internet, it is difficult to establish whether the infinitive must be analyzed as an impersonal verb or not. In
addition, according to the native speakers I consulted, this sentence is not acceptable, or at least very
marked.
41
This article has provided an analysis of the restriction on infinitives of impersonal verbs in
Russian. In contrast to the existing analysis by Perlmutter and Moore (2002), and in the same
vein as the formal analysis given by Babby (2009), I have argued that this restriction is not
connected to case assignment. Instead, I have shown that the data can be explained with
reference to various morphological, syntactic and semantic factors.
The infinitive is not morphologically marked for its impersonal status, and the
impersonal status of the infinitive is only triggered – at least in some syntactic contexts – if
there is another impersonal verb present in the context. If no other impersonal verb is present,
the subject of the infinitive is interpreted as a human (generic or arbitrary) subject. Only if the
infinitive occurs as the complement of an impersonal auxiliary verb is the infinitive interpreted
as an impersonal verb. This results from the fact that the situation expressed by the auxiliary and
the situation expressed by the infinitive together express a complex single event, as well as form
a syntactic unit. If the infinitive occurs with a subordinator, the infinitive is strongly associated
with an agentive subject, and the interpretation of the infinitive as an impersonal is blocked.
However, with some impersonal verbs, particularly meteorological verbs, and in some
constructions, particularly constructions that do not presuppose human or agentive subjects,
impersonal infinitives are (sometimes marginally) acceptable. Instances such as these suggest
that the underlying explanatory factor is not case assignment, and that both syntactic and
semantic factors play a part in the interpretation of the data.
It is my belief that the explanation I have given here is formulated in such a way that it
can be invalidated by new data, and, in the same vein, can make predictions with respect to new
data. An example of a potential contradiction to my explanation would be a subordinator that is
not associated with agentive semantics but nonetheless blocks infinitives of meteorological
impersonals. Since I have given a fairly complete overview of the Russian data, the existence of
this potential contradiction is, of course, largely hypothetical. Note, however, that my
explanation also makes predictions about infinitives in other languages. On the basis of my
analysis one can predict the occurrence of similar restrictions on infinitives of impersonals,
42
provided that the infinitive is associated with agentive semantics, as in Russian. In this analysis
I have only briefly compared Russian to Italian and Dutch, and indicated that Dutch shows
similarities to Russian with respect to the use of infinitives of impersonals, but a typological
analysis would be a welcome addition to my interpretation. In addition, some claims I have
made could be corroborated or challenged by psycholinguistic research. Psycholinguistic
research could investigate, for example, whether or not the claim that Russian infinitives are
easily associated with human subjects or agentive features is true. I leave these issues for further
research.
References
Babby, L.H., 2009. The syntax of argument structure. Cambridge University Press, Cambridge.
Babby, L.H., 2010a. Voice and Diathesis in Slavic. Glossos 10.
http://www.seelrc.org/glossos/issues/10/babby.pdf
Babby, L.H. 2010b. Prolegomenon to any future typology of impersonal sentences in Russian. In
Hypothesis A and Hypothesis B: Papers in Honor of David Perlmutter, D. Gerts, J. Moore &
M. Polinsky (eds), 19–40. Cambridge MA: The MIT Press.
Birjulin, L., 1993. Sintaktičeskoe predstavlenie russkogo impersonala (tipa svetaet): K istorii i teorii
voprosa. Russian Linguistics 17, 181–203.
Bricyn, V., 1990. Sintaksis i semantika infinitiva v sovremennom russkom jazyke. Naukova
Dumka, Kiev.
Comrie, B., 1974. The second dative: a transformational approach. In: Brecht, R., Chvany,
C (Eds.), Slavic Transformational Syntax. Ann Arbor, University of Michigan, pp. 123–150.
Ebeling, C., 1984. On the meaning of the Russian infinitive. In: van Baak, J. (Ed.), Signs of
friendship. Rodopi, Amsterdam, pp. 97–130.
Egerland, V., 2003. Impersonal pronouns in Scandinavian and Romance. Working papers in
Scandinavian syntax 71, 75–102.
43
Fortuin, E.L.J., 2003. The conceptual basis of syntactic rules: a semantic motivation for the second
dative in Russian. Lingua, an International Review of General Linguistics 113, 49-92.
Fortuin, E., 2006. The use of dative subjects in the conjunction of anteriority in Russian. Russian
Linguistics 30 (3), 321–357.
Fortuin, E. 2005. From possibility to necessity: the semantic spectrum of the dative-infinitive
construction in Russian". In: Hansen, B. & Karlík, P. (eds.), Modality in Slavonic Languages.
New Perspectives. München: Otto Sagner, pp. 39-60.
Franks, S., 1990. Case, configuration, and argumenthood: Reflections on the second dative.
Russian Linguistics 14, 231–254.
Franks, S., 1995. Parameters of Slavic morphosyntax. Oxford University Press, New York.
Guiraud-Weber, M., 1984. Les Propositions sans nominative en Russe moderne. Institut
d’études Slaves, Paris.
Keenan, E. 1976. Towards a universal definition of ‘subject’. In: Li, Ch. (Ed.), Subject and
Topic. NY/SF/London, pp.303-333.
Moore, J., Perlmutter, D.M., 2000. What does it take to be a Dative Subject? Natural Language
and Linguistic Theory 18, 373–416.
Ožegov, S. I. 1973. Slovar’ Russkogo Jazyka. 10-e izdanie. Moskva.
Perlmutter, D., 2007. In what ways can finite and nonfinite clauses differ? Evidence from
Russian. In: Nikolaeva, I. (Ed.), Finiteness: Theoretical and Empirical Foundations. Oxford
University Press, Oxford, pp. 250–305.
Perlmutter, D., Moore, J., 2002. Language-internal explanation: the distribution of Russian
Impersonals. Language 78: 619–650.
Rubinstein, G., 1984. Subjective dative in Russian infinitival clauses of purpose. Slavic and
East European Journal 30, 367–379.
Von Fintel, K., Iatridou, S., 2007. Antonomy of a modal construction. Linguistic Inquiry 38,
445–483.
44
Appendix
Frequencies and statistics
Sentence number
Mean
Median
Std Deviation
(7)
3,53
4,00
1,231
Finite counterpart (7)
4,19
5,00
1,115
(8)
3,31
4,00
1,259
Finite counterpart (8)
2,55
2,00
1,300
(9)
1,79
1,00
1,196
Finite counterpart (9)
3,17
4,00
1,403
(10)
2,60
2,00
1,401
(38)
3,57
4,00
1,313
(11)
2,16
2,00
1,073
Finite counterpart
2,98
3,00
1,207
(32)
2,43
2,00
1,117
(33)
2,60
2,00
1,401
(55)
3,91
4,00
1,204
Finite counterpart
3,14
3,00
1,432
(56)
3,29
4,00
1,228
(57)
2,29
2,00
1,185
(58)
2,55
2,00
1,259
(11)
(55)
45